rsos.royalsocietypublishing.org

ResearchCite this article:Wang Y, Pham DT, Zhang Z,Li J, Ji C, Liu Y, Leng J. 2016 Sustainableself-healing at ultra-low temperatures instructural composites incorporating hollowvessels and heating elements. R. Soc. open sci.3: 160488.http://dx.doi.org/10.1098/rsos.160488

Received: 13 July 2016Accepted: 17 August 2016

Subject Category:Engineering

Subject Areas:mechanical engineering/materials science

Keywords:carbon nanotubes, self-healing,fibre-reinforced composites, delamination,self-repair, smart materials

Author for correspondence:Yongjing Wange-mail: yxw181@bham.ac.uk

Electronic supplementary material is availableat http://dx.doi.org/10.1098/rsos.160488 or viahttp://rsos.royalsocietypublishing.org.

Sustainable self-healing atultra-low temperatures instructural compositesincorporating hollowvessels and heatingelementsYongjing Wang1, Duc Truong Pham1, Zhichun Zhang2,

Jinjun Li4, Chunqian Ji1, Yanju Liu3 and Jinsong Leng2

1Department of Mechanical Engineering, School of Engineering, University ofBirmingham, Edgbaston, Birmingham, UK2Center for Composite Materials and Structures, and 3Department of AerospaceScience and Mechanics, Harbin Institute of Technology, Science Park, Harbin,People’s Republic of China4Applied Science Faculty, Delft University of Technology, Delft, The Netherlands

YW, 0000-0002-9640-0871

Self-healing composites are able to restore their propertiesautomatically. Impressive healing efficiencies can be achievedwhen conditions are favourable. On the other hand, healingmight not be possible under adverse circumstances such asvery low ambient temperature. Here, we report a structuralcomposite able to maintain its temperature to provide asustainable self-healing capability—similar to that in thenatural world where some animals keep a constant bodytemperature to allow enzymes to stay active. The compositeembeds three-dimensional hollow vessels with the purposeof delivering and releasing healing agents, and a porousconductive element to provide heat internally to defrost andpromote healing reactions. A healing efficiency over 100%at around −60°C was obtained. The effects of the sheets onthe interlaminar and tensile properties have been investigatedexperimentally. The proposed technique can be implementedin a majority of extrinsic self-healing composites to enableautomatic recovery at ultra-low temperatures.

1. IntroductionSelf-healing composite materials are artificial materials that canheal after damage like living creatures do. Research efforts

2016 The Authors. Published by the Royal Society under the terms of the Creative CommonsAttribution License http://creativecommons.org/licenses/by/4.0/, which permits unrestricteduse, provided the original author and source are credited.

on May 22, 2018http://rsos.royalsocietypublishing.org/Downloaded from

2

rsos.royalsocietypublishing.orgR.Soc.opensci.3:160488

................................................across the globe over the past two decades have resulted in healing efficiencies above 100%, indicatingthat the function or performance of the healed material can be better than that prior to damage [1]. Suchhealing is expected to be of great value in applications where repairing or replacing key components ischallenging or even impossible, for instance, in aircraft and satellites during flight, and in equipmentthat is difficult to access such as offshore wind turbines.

Fibre-reinforced composites (FRCs) have gained popularity in the above-mentioned applications dueto their high strength and light weight. However, the main risk in employing these materials is internalmicro-cracks which may cause catastrophic failures and that are hard to detect and repair. Hence,enabling them to self-heal has been proposed as a potential method to improve the reliability of FRCsand increase their service life as well as to decrease repair costs. To enable a composite material toself-heal, capsule-based [2] and vessel-based designs [3] have been proposed. A capsule-based designinvolves embedding capsules containing special liquids able to cause healing of the host material. Whena crack occurs, some of the capsules break and release the liquids (also known as healing agents)which fill the crack. A vessel-based design works in a similar way, but the capsules are replaced bya vascular network in one, two or three dimensions. By adopting these designs, FRCs endowed withthe ability to self-heal have been created. Blaiszik et al. [4] and Jones et al. [5,6] have developed fibresfunctionalized with micro-capsules containing healing liquids to repair the interface between the fibresand the polymer matrix [4–6]. Moll et al. [7] dispersed the capsules in the matrix and reported a near100% recovery from minor damage. Norris et al. [8] embedded one-dimensional hollow vessels loadedwith healing agents in an FRC. Patrick et al. [9] invented an FRC incorporating three-dimensionalvessels in a complex pattern to cope with delamination and restore mechanical properties in multipledamage cycles.

However, it is worth noting that the good healing performances reported so far are attainableonly when there are favourable healing conditions, such as a suitably high ambient temperature oran appropriate radiation treatment, which is often impossible to ensure in practical applications. Forinstance, composites used on an aircraft may endure temperatures as low as −60°C, at which almostall healing liquids would be frozen and cannot be activated. This has become one of the main barriersto the wider adoption of self-healing composites, prompting efforts to develop systems that can self-heal regardless of environmental and damage conditions. A few researchers have been trying to do thisby employing new healing liquids, and have reported healing agents able to heal at temperatures aslow as 10°C [10,11]. The effects of ultra-low temperatures on a typical healing agent have also beeninvestigated [12]. However, real high-efficiency healing at very low temperatures (−40 to −80°C) isstill impossible.

Here, we report a design to enable self-healing in FRCs at ultra-low temperatures. With theproposed approach, healing was fulfilled by two components: three-dimensional vessels and athin layer of conductive material. The vessels were embedded inside the structural compositeswith the purpose of delivering and releasing healing agents. The thin layer was to supply heatinternally from the composites to cause de-icing and provide a suitable temperature for healing.We selected a vessel-based design instead of a capsule-based design because the former is capableof recovery from large-area damage, as healing agents can be continuously pumped into thevessels [3,9,13–18]. Furthermore, the vessel network would in general only have minor effectson the tensile properties of the composites [19]. The fabricated composites are able to recoverfrom severe delamination with average efficiencies around 100% at ultra-low temperatures. Wealso discuss the effects of the conductive sheets on interlaminar and tensile properties of thelaminates. Experimental results indicate that the sheets reduced interlaminar strength but increasedtensile properties.

2. Material and methods2.1. Structure of the compositeThe composite is a glass fibre-reinforced laminate embedded with wave-like hollow vessels andconductive sheets, as shown in figure 1a. When delamination occurs at low temperature, cracks canpropagate and break the vessels. Heat could be generated internally through electrical heating to defrost,releasing liquid healing agents into the cracks. Solidification of the agents could also be accelerated by theinternal heat. The healing process is shown in figure 1b. Here, we discuss the wave-like hollow vesselsand conductive sheets separately.

on May 22, 2018http://rsos.royalsocietypublishing.org/Downloaded from

3

rsos.royalsocietypublishing.orgR.Soc.opensci.3:160488

................................................

carbonnanotubesporous sheet(or copperfoam sheet)

the compositesheal throughreleasinghealing agentand internalheating

delaminationoccurs causingsome of thevessels tobreak

multiplereinforcementlayers

core of thecompositeembedded withvessels

vessels filledwith healing

agent

(a) (b)

Figure 1. (a) Internal structure of the composites and (b) damage–bleeding–healing process.

2.1.1. Wave-like hollow vessels in the core

The wave-like configuration was proposed by Patrick et al. [9] who were the first to use it as well as aherringbone configuration for self-healing in FRCs. The hollow vessels can be made by the vapourizationof sacrificial components (VaSC) incorporated in the material with the reinforcement fibres. Poly(lacticacid) (PLA) sacrificial fibres 300 µm in diameter were selected as sacrificial components as they leftlittle residue, greatly reducing the risk of blocking the channels [9,20–22]. Other methods to fabricateinternal hollow structures include using hollow fibres [23,24], electrostatic discharging [25] and laserdirect-writing [26]. However, only VaSC can produce large-scale three-dimensional vascular networksfollowing an accurate pre-designed pattern.

2.1.2. Conductive porous sheets

The conductive layer must satisfy two requirements: good electrical conductivity and good thermalconductivity. To fabricate this conductive layer, two types of conductive sheet were selected: porouscopper foam sheet (CFS) and porous carbon nanotube sheet (CNS). Metal foam has already been appliedin batteries [27], heat exchange devices [28] and energy absorbers [29]. The CFS (figure 2a) had a thicknessof 0.5 mm and a porosity of 96 ∼ 98%. It had a thermal conductivity around 10 W m−1 K−1, very highelectrical conductivity (approx. 3.9 × 106 S m−1) and a large area of contact with the host material. Theother conductive sheet (figure 2b), CNS, with a thickness of 40 µm, had a stable electrical conductivity of1.25 ∼ 1.38 × 10−4 Ω m and a thermal conductivity estimated to be in the range 100 ∼ 1000 W m−1 K−1.The thermal conductivity was measured by using the method in the work of Wang et al. [30]. The CNSwas fabricated by multiple steps of single wall carbon nanotubes (CNTs) dispersion and suspensionfiltration [31].

2.2. Fabrication procedureVaSC and resin infusion were used to make FRCs incorporating hollow vessels and heating components.The PLA sacrificial fibres (300 µm VascTech fibres, CU Aerospace Ltd.) were manually embedded intoeight layers of woven glass fibres (area density of 290 g m−2 for each layer) in a square-wave-likeconfiguration. The incorporated reinforcement fibres and other untreated reinforcement fibres, as wellas the conductive sheets, were deposited layer by layer in the following sequence: bottom—four layersof normal glass fibres—eight layers of glass fibres with the sacrificial components—two layers of normal

on May 22, 2018http://rsos.royalsocietypublishing.org/Downloaded from

4

rsos.royalsocietypublishing.orgR.Soc.opensci.3:160488

................................................

copper foam sheet carbon nanotube sheet SEM

(b)(a)

Figure 2. (a) Porous copper foam sheet and (b) porous carbon nanotube sheet.

glass fibres—conductive sheet—two layers of normal glass fibres—top. A nylon sheet was placed atthe mid-plane position and offset 30 mm from one of the edges. The sheet served as a crack createdduring the fabrication of the composite. Epoxy resin and hardener (very high temperature epoxy, EasyComposites Ltd.) were mixed at a ratio of 100 : 35 parts by weight and degassed at 35°C for 30 min in avacuum chamber. Afterwards, a resin infusion process took place to make the composites. The mixturewas cured for 36 h at room temperature and then put through post-cure heating cycles at 40°C, 60°C,80°C, 100°C and 120°C each for 1 h, and 140°C for 3 h, as suggested by the supplier. After the resin wasfully cured, the composite which had a thickness of 4 mm was cut into 180 × 25 mm pieces using a gritsaw and polished with sand paper. The cross section of the samples is shown in figure 2a (FRCs + CFS)and 2b (FRCs + CNS).

The sacrificial components were removed by placing the samples in a 200°C vacuum chamber for 24 h.After the fabrication of the hollow vessels, the healing agent, which was a pre-mixed two-part epoxy(RT151, ResinTech Ltd.) dyed in red, was injected into the vessels using a controllable liquid dispenser.The complete fabrication procedure is detailed in the electronic supplementary material.

2.3. Healing performance assessment and analysisThe mechanical strength of the samples was assessed using the double cantilever beam (DCB) test, asshown in figure 3a, following the procedure described in §2.5. For each sample, a total of three DCBtests were conducted. The first test was to measure the original interlaminar strength and to breakthe sample. Afterwards, the broken sample was loaded again for the second test to reveal the residualstrength. Then it was given a 24 h rest to heal at −60°C before taking the third test to assess the healingperformance. The healing process started with a continuous injection of healing agents into the vessels,followed by the specimens being placed in an ultra-low temperature (−60°C) chamber. During healing,the conductive layer was electrically heated. The power used was 7 W and 9.45 W for CFS and CNS,respectively. A thermometer was attached to the specimen to monitor its temperature. The cross sectionof the healed sample is shown in figure 3b. During the test, cracks propagated and traversed the wave-likevessels as in figure 3c.

2.3.1. Calculation of healing efficiency

The healing efficiency in relation to peak load (η1) was calculated as [9,32]:

η1 = LHealed

LVirgin× 100%, (2.1)

on May 22, 2018http://rsos.royalsocietypublishing.org/Downloaded from

5

rsos.royalsocietypublishing.orgR.Soc.opensci.3:160488

................................................(a) (c)

(b)

compositesincorporating

CFS

compositesincorporating

CNS

CNS

original after healing

hollowvessel

delamination results from

DC

B test

Figure 3. (a) DCB test of a healed composite specimen incorporating CNS, (b) composites incorporating CNS and CFS and (c) fibre-reinforced composite incorporating wave-like micro vessels after DCB test.

where LVirgin is the achieved maximum load before damage and LHealed is the maximum load after thespecimen has recovered from an 80 mm mode-I fracture.

The healing efficiency in relation to fracture energy (η2) was calculated as [9]:

η2 = UI,Healed

UI,Virgin× 100%, (2.2)

where UI,Healed and UI,Virgin are the fracture energies of a specimen before damage and after recovery,respectively, derived from first principles as the area under the load–displacement trace at a particularcrack length.

2.4. Effects of the conductive sheets on interlaminar propertiesThe introduction of a conductive sheet created a new layer inside the laminates. The effects of having thislayer on interlaminar properties were revealed using DCB tests to compare samples with and withoutthe sheets.

Three groups of samples, each containing five specimens, were made using the techniques describedin §2.2. In the first group, specimens were fibre-reinforced laminates with the CNS in the middle layer. Inthe second group, the CNS was replaced by the CFS. Specimens in the third group were ordinary fibre-reinforced laminates. All specimens were 120 mm in length, 20 mm in width and 4 mm in thickness, andhad a 20 mm deep pre-crack on one edge. DCB tests followed the procedure outlined in §2.5.

2.5. Double cantilever beam testAluminium piano hinges were attached onto the specimen using a structural adhesive (Loctite 330 Glueand Activator Multibond Kit). After heating for 6 h at 60°C, the adhesive was fully cured and testing wascarried out on an MTS Criterion Model 43 machine. The specimens were loaded through the bondedhinges in quasi-static tension to induce mode-I fracture propagation along the mid-ply interlaminarregion until the crack reached 80 mm. The displacement-controlled crosshead speed was 5 mm min−1

during loading.

on May 22, 2018http://rsos.royalsocietypublishing.org/Downloaded from

6

rsos.royalsocietypublishing.orgR.Soc.opensci.3:160488

................................................

matrixE1

E2 C2

C1

reinforcementfibres

CNT sheet

CNT sheet

CNT sheet

random-discontinuouscotton fibre

woven carbonfibres

(a) (b)

(c) (d )

(e)

Figure 4. (a) Schematic of composite E, (b) schematic of composite C, (c) CNT porous sheet, (d) random-discontinuous cotton fibres andCNT sheet and (e) woven carbon fibres and CNT sheet.

Table 1. Details of the subgroups of composites.

subgroup code description no. specimens

E1 random-discontinuous cotton fibre composites 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

E2 random-discontinuous cotton fibre composites embedded with porous CNT layer 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

C1 woven carbon fibre composites 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

C2 woven carbon fibre composites embedded with porous CNT layer 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2.6. Effects of the conductive sheets on tensile propertiesThe effects on two types of composites were investigated: random-discontinuous cotton-FRCs (type E)and woven carbon-FRCs (type C) (figure 4). E contained four layers of cotton breather cloth and C hadfour layers of woven carbon fibres. As the fillers are different, E and C have different bearing strengths,representing weak and strong polymer composites, respectively.

E and C are further divided into two subgroups (E1/E2 and C1/C2) as shown in table 1. The effectsof the CNT layer can be revealed through comparing E1 (or C1) with E2 (or C2).

The porous CNT layer was fabricated inside E2 and C2 by embedding a CNS using the techniquesdescribed in §2.2. After the resin was fully cured, the composites were cut into 100 × 9 × 2 mm specimens.The specimens were placed in a chamber at 50°C for 12 h to release internal strain. A MTS Criteriontesting machine (MTS Criterion Model 43 Electromechanical Universal Test System) was used to collectdata. Each specimen was subject to tension parallel to its long edge whose effective length is 15 mm.The loading rate was 0.5 mm min−1. After tensile testing, the samples were observed under optical andelectron microscopy to reveal cracking patterns and the nature of the fracture.

3. Results and discussion3.1. De-icing performanceFigure 5a shows the temperature of a sample incorporating CNS under electrical heating. With theapplied voltage maintained between 10 and 16 V, the steady-state temperatures of the specimens

on May 22, 2018http://rsos.royalsocietypublishing.org/Downloaded from

7

rsos.royalsocietypublishing.orgR.Soc.opensci.3:160488

................................................

100

tem

pera

ture

(°C

)

80

60

40

20

0

–20

–40

–600 200 400 600

10 V12 V14 V16 V

frozen area de-iced area

T = 0 s

T = 20 s

T = 30 s

T = 40 s

T = 80 s

800 1000 1200 1400time (s)

(a)

(b) (c)

Figure 5. (a) Temperature of the specimen as a function of time and applied voltage, (b) thermal distribution in the composite heatedby CNS in an ultra-low temperature chamber and (c) de-icing of composite with CNS.

remained in the range 20–85°C (figure 5a), which was sufficiently high to enable curing in 24 h (figure 5b).There was no severe heat concentration due to the use of a relatively low electrical current (200 mA) andits good thermal conductivity. The composite was able to be fully de-iced in 90 s (figure 5c), which alsodemonstrates the efficiency of CNS.

The composite with CFS was able to keep the temperature in the range 5–20°C. However, for thedesign using CFS, it was found that although the healing agent was still active in the temperaturerange, 24 h was not long enough for it to cure fully. Increasing the electrical power to raise the internaltemperature may seem a solution to this problem. However, this was almost impossible as the copperfoam had an extremely low resistivity, and so a very large electrical current—55 A in this case—wasrequired to generate the heat necessary to keep the healing agent active. Such a high electrical currentmight generate extremely high temperature at the points of contact between the sample and the powersupply and may cause local melting. In our experiments, although four of the five samples testedsuccessfully healed themselves, there was still one that failed due to heat concentration. Therefore,although increasing the supplied power will raise the internal temperature, it also heightens the riskof local overheating as a side effect.

3.2. Healing performanceAfter 24 h of healing at −60°C, the recovered mechanical properties of the samples are given in tables 2and 3 and figure 6a,b.

For the composite with CNS, an average healing efficiency of 107.7% for fracture energy and 96.22%for peak load was achieved. The maximum healing efficiency for fracture energy was 141%.

Four samples of the composites with CFS, out of five, successfully recovered from damage with anaverage healing efficiency of 63.2% for fracture energy and 58.8% for peak load. The maximum healingefficiency for fracture energy was 128%.

on May 22, 2018http://rsos.royalsocietypublishing.org/Downloaded from

8

rsos.royalsocietypublishing.orgR.Soc.opensci.3:160488

................................................

120

40

108%

63.20%

originaldamagedhealing without CNShealing with CNS

healing efficiency

strain energy peak load

96%

58.80%

100

80

(%)

60

40

20

00 20 40

displacement (mm)60 80

5

10

15

20

load

(N

) 25

30

35with CFSwith CNS

(a) (b)

Figure 6. (a) Healing efficiency of composite with CNS and CFS and (b) displacement–load curve for a typical sample incorporating CNS.

Table 2. DCB test results quantifying healing performances.

specimen no.

originalpeak load(N)

recoveredpeak load(N)

originalfracture energy(N mm)

recoveredfracture energy(N mm)

healingefficiency forpeak load (%)

healing efficiencyfor fractureenergy (%)

GFRC+ CNS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 31.5 32.5 1544.2 2077.4 103 135. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 30 29.8 1437.5 728.8 99.30 50.70. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3 31.2 26.6 1356.8 1329 85.30 98. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4 28 26.44 1586.5 2238 99.40 141. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5 32.3 30.4 1565.35 1783.5 94.10 114. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

GFRC+ CFS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 65 40 3484.8 3298.4 61.50 94.70. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 62 45 2978.6 1837.3 72.60 61.70. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3 58 30 3610.2 1141.7 51.70 31.60. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4 51 55 2564.3 3284.1 108 128. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5 55 n.a. 3024.5 n.a. 0 0. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Table 3. Healing efficiency summary.

no. samples

max. healingefficiency forfractureenergy (%)

min. healingefficiency forfractureenergy (%)

ave. healingefficiency forfractureenergy (%)

max. healingefficiency forpeak load(%)

min. healingefficiency forpeak load(%)

ave. healingefficiency forpeak load(%)

GFRC+ CNS 5 141 50.7 107.7 103 85.3 96.22. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

GFRC+ CFS 5 128 0 63.2 108 0 58.8. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

The results indicate that the composite materials with either CFS or CNS are able to self-heal at ultra-low temperatures. Furthermore, CNS is the better of the two types of conductive layer. The electricalresistivity of CFS was too low and therefore a very high electrical current was required to generate heat,causing potential overheating for everything connected in series with the composite, especially at thecontact points. Such a high current may locally generate temperatures approaching 500°C. Also, thethermal conductivity of CFS is not high enough to allow heat transfer to other areas. This caused partsof the composite to be damaged by heat concentration, while the rest remained so cold that the healingagents could not function properly. As a result, the composite only achieved an average healing efficiencyno greater than 70%. CNS, on the other hand, had an electrical resistivity which was much higher than

on May 22, 2018http://rsos.royalsocietypublishing.org/Downloaded from

9

rsos.royalsocietypublishing.orgR.Soc.opensci.3:160488

................................................

control

control

strain energy (N mm)1800

0.14

peak load (N)

CFS

CFS

CNS

CNS

control CFS CNS0

140

120

100

80

60

40

20

0

0 40353025displacement (mm)

2015105

0 403530252015105

0 403530252015105

200400600800

1000120014001600

00.020.040.060.08

load

(kN

)lo

ad (

kN)

load

(kN

)0.100.12

0.14

00.020.040.060.080.100.12

0.14

00.020.040.060.080.100.12

(a)

(b)

(c)

(d)

(e)

Figure 7. Mode I load versus displacement curve for the laminates with (a) no embedment, (b) a copper foam sheet and (c) a carbonnanotube sheet. Results of DCB testing: (d) fracture energy and (e) peak load.

that of CFS. Thus, only a small electrical current was sufficient to generate the required heat. The thermalconductivity was also sufficiently high to enable heat to flow rather than concentrate at local hot spots.

To achieve satisfactory healing performances, another important factor was how well the healingagent covers the composite materials. There have been concerns that the wave-like design in which thevessels penetrate multiple layers cannot ensure a large coverage and might be less effective than thatadopted in straight hollow-fibre-based self-healing composites [24,33,34]. However, the high healingefficiency proved that a large coverage was achievable. Another ongoing study by the authors hasshown that having a large coverage of vessels in multiple layers is more important than having extensiveinterlaminar coverage [35].

It might be possible to achieve self-healing at ultra-low temperatures using new healing agents.However, it is worth noting that any healing agent will have an active temperature range, outsideof which the healing process cannot take place. All established healing agents so far have narrowactive temperature ranges, and none of them is suitable when a high healing efficiency over a widetemperature range, such as −60 to 100°C, is required. Even if a new healing agent can be developed forthat temperature range, it would become ineffective as soon as the operating temperature fell outsidethe range. Thus, the development of new healing agents is not an effective way to achieve sustainableself-healing. The essence of our approach is that there is no limitation relating to types of healing agentsand composites—all established composites that self-heal in certain temperature ranges can be modifiedbased on the design presented here to achieve self-healing at ultra-low temperatures. By tuning thepower provided to the conductive layer, the internal temperature of composites can be altered regardlessof the ambient temperature to provide suitable conditions for any healing agent.

3.3. Effects of conductive sheets on interlaminar propertiesThe mode I load versus displacement curves for composites with and without conductive sheets areshown in figure 7a–c. Each figure displays the load–displacement curves for five tested specimens.Comparison of composites with and without conductive sheets (figure 7d,e) indicates obvious reductions

on May 22, 2018http://rsos.royalsocietypublishing.org/Downloaded from

10

rsos.royalsocietypublishing.orgR.Soc.opensci.3:160488

................................................

cotton fibre

composites

peak stress (MPa)

10 000

8000

6000

4000

2000

0

3000

2500

2000

1500

1000

500

0without CNS with CNS

strain at break (mm mm–1) modulus (MPa)

woven

carbon fibre

composites

without CNS with CNS without CNS with CNS

without CNS with CNS without CNS with CNS without CNS with CNS

80

1000

800

600

400

200

0

706050403020100

0.06

0.05

0.04

0.03

0.02

0.01

0

0.5

0.4

0.3

0.2

0.1

0

Figure 8. Tensile properties of composites with and without carbon nanotube sheets.

(b)(a)

(c) (d ) (e)

( f ) (g)

Figure 9. Scanning electron microscopy (SEM) images of carbon fibre composite incorporating carbon nanotube sheets after tensiletests. SEM image of crack in carbon fibre-reinforced composite incorporating carbon nanotube sheet. (a,b) Cross section of a damagedsample, (c–e) delamination of carbon fibres and matrix, (f,g) delamination of carbon nanotube sheet and matrix.

in fracture energy and peak loads. The adoption of a porous sheet was to form a short-fibre-reinforcedlayer inside the composite so that the layer could bond strongly to the host material. The results mightbe explained by the possibility that the resin had not fully infiltrated the porous layer and that theconductive sheet should have a very rough surface to strengthen the interface bonding with the hostmaterial. A potential solution might be to optimize the surface pattern of the sheet to help preserve the

on May 22, 2018http://rsos.royalsocietypublishing.org/Downloaded from

11

rsos.royalsocietypublishing.orgR.Soc.opensci.3:160488

................................................interlaminar strength while ensuring heating efficiency. Techniques similar to those used with vascularnetworks for vessel-based self-healing materials could be adopted as the same potential issue withdelamination exists there also [36–38].

3.4. Effects of carbon nanotube sheets on tensile propertiesThe tensile properties of composites with and without the carbon nanotube sheets are shown infigure 8. The experimental results indicate that CNTs can significantly improve the tensile strength ofpolymer composites without much affecting their elasticity modulus. For cotton fibre composites, failurehappened with fracture of both the CNT sheet and the surrounding material and there was no obviousdelamination. Considering that the CNT sheet was the primary loading component in cotton fibrecomposites, it can be expected that they failed once the sheet had reached its ultimate strain and ruptured.Unlike in cotton fibre composites, failure in carbon fibre composites was the result of a combination ofrupture of the sheet, breaking of reinforcement fibres and delamination, as shown in figure 9. However, ina carbon fibre composite material, delamination always needs to propagate through the matrix betweentwo adjacent carbon fibre layers. When a layer of CNTs is present, a crack would need to penetrateit before reaching the next carbon fibre layer and this provides extra resistance to crack propagationand delamination. This could be the reason why the introduction of CNT sheets improved the tensileproperties of carbon fibre composites.

4. ConclusionWe have shown that self-healing at ultra-low temperatures can be implemented by adding vessels anda porous conductive layer into a composite material. Healing agents were continuously injected intothe vessels and were released after the host material was damaged. The conductive layer increasedthe temperature of the composites through electrical heating to assist the flow and curing of healingagents. Both the CFS and the CNS were able to act as conductive layers. However, the composite withthe CNS was able to self-heal more effectively and stably. As a result, healing in FRCs at a temperaturearound −60°C was achieved with an average recovery of 107.7% in fracture energy and 96.22% inpeak load. The effects of the conductive sheet on interlaminar properties and tensile properties wereexperimentally investigated. It was found that the introduction of a CNS increased the tensile strength ofpolymer composites, but had negative effects on interlaminar properties. A potential solution might be toinclude patterns on the sheet surface to preserve interlaminar properties without significantly affectingheating performance.

Data accessibility. The methods and materials used in this work are described in the body of the manuscript and theelectronic supplementary material.Authors’ contributions. Y.W. contributed to the initial conception, participated in the experimental work, theoretical andexperimental data analysis, results presentation and interpretation and drafted the manuscript. D.T.P. supervised theresearch and contributed to the interpretation of experimental results and manuscript writing. Z.Z. participated infabrication of the CNS and the acquisition of data. J.L. participated in the experimental work and the analysis of data.C.J. contributed to the initial conception and manuscript revisions. Y.L. and J.L. coordinated the manufacture of theCNS and advised on experimental work. All authors gave final approval for publication.Competing interests. We declare we have no competing interests.Funding. The research was partially supported by TSB and EPSRC (grant no. EP/L505225/1) and Research Exchangewith China and India, The Royal Academy of Engineering, UK (grant no. 1415-1).Acknowledgements. The authors acknowledge the EPSRC Engineering Instrument Pool for the loan of instruments. Weare thankful to Mr Carl Hingley, Mr Peter Thornton and Mr Hengkun Liang for their help with experiment designand preparations.

References1. Wang Y, Pham DT, Ji C. 2015 Self-healing

composites: a review. Cogent Eng. 2, 1075686.(doi:10.1080/23311916.2015.1075686)

2. White SR, Sottos NR, Geubelle PH, Moore JS, KesslerMR, Sriram SR, Brown EN, Viswanathan S. 2001Autonomic healing of polymer composites. Nature409, 794–797. (doi:10.1038/35057232)

3. Toohey KS, Sottos NR, Lewis JA, Moore JS, White SR.2007 Self-healing materials with microvascularnetworks. Nat. Mater. 6, 581–585. (doi:10.1038/nmat1934)

4. Blaiszik BJ, Baginska M, White SR, Sottos NR. 2010Autonomic recovery of fiber/matrix interfacialbond strength in a model composite. Adv. Funct.

Mater. 20, 3547–3554. (doi:10.1002/adfm.201000798)

5. Jones AR, Blaiszik BJ, White SR, Sottos NR. 2013 Fullrecovery of fiber/matrix interfacial bond strengthusing a microencapsulated solvent-based healingsystem. Compos. Sci. Technol. 79, 1–7. (doi:10.1016/j.compscitech.2013.02.007)

on May 22, 2018http://rsos.royalsocietypublishing.org/Downloaded from

12

rsos.royalsocietypublishing.orgR.Soc.opensci.3:160488

................................................6. Jones AR, Cintora A, White SR, Sottos NR. 2014

Autonomic healing of carbon fiber/epoxyinterfaces. ACS Appl. Mater. Interface 6, 6033–6039.(doi:10.1021/am500536t)

7. Moll JL, Jin H, Mangun CL, White SR, Sottos NR. 2013Self-sealing of mechanical damage in a fully curedstructural composite. Compos. Sci. Technol. 79,15–20. (doi:10.1016/j.compscitech.2013.02.006)

8. Norris CJ, Meadway GJ, O’Sullivan MJ, Bond IP, TraskRS. 2011 Self-healing fibre reinforced composites viaa bioinspired vasculature. Adv. Funct. Mater. 21,3624–3633. (doi:10.1002/adfm.201101100)

9. Patrick JF, Hart KR, Krull BP, Diesendruck CE, MooreJS, White SR, Sottos NR. 2014 Continuousself-healing life cycle in vascularized structuralcomposites. Adv. Mater. 26, 4302–4308.(doi:10.1002/adma.201400248)

10. Schunack M, Gragert M, Döhler D, Michael P, BinderWH. 2012 Low-temperature Cu(I)-catalyzed ‘click’reactions for self-healing polymers.Macromol.Chem. Phys. 213, 205–214. (doi:10.1002/macp.201100377)

11. Hillewaere XKD, Du Prez FE. 2015 Fifteen chemistriesfor autonomous external self-healing polymers andcomposites. Prog. Polym. Sci. 49–50, 121–153.(doi:10.1016/j.progpolymsci.2015.04.004)

12. Raimondo M, Longo P, Mariconda A, Guadagno L.2015 Healing agent for the activation of self-healingfunction at low temperature. Adv. Compos. Mater.24, 519–529. (doi:10.1080/09243046.2014.937135)

13. White SR, Moore JS, Sottos NR, Krull BP, Cruz WAS,Gergely RCR. 2014 Restoration of large damagevolumes in polymers. Science 344, 620–623.(doi:10.1126/science.1251135)

14. Hamilton AR, Sottos NR, White SR. 2012 Pressurizedvascular systems for self- healingmaterials. J. R. Soc.Interface 9, 1020–1028. (doi:10.1098/rsif.2011.0508)

15. Hansen CJ, White SR, Sottos NR, Lewis JA. 2011Accelerated self-healing via ternaryinterpenetrating microvascular networks. Adv.Funct. Mater. 21, 4320–4326. (doi:10.1002/adfm.201101553)

16. Hamilton AR, Sottos NR, White SR. 2010Self-healing of internal damage in syntheticvascular materials. Adv. Mater. 22, 5159–5163.(doi:10.1002/adma.201002561)

17. Hansen CJ, WuW, Toohey KS, Sottos NR, White SR,Lewis JA. 2009 Self-healing materials withinterpenetrating microvascular networks. Adv.Mater. 21, 4143–4147. (doi:10.1002/adma.200900588)

18. Toohey KS, Hansen CJ, Lewis JA, White SR, SottosNR. 2009 Delivery of two-part self-healingchemistry via microvascular networks. Adv. Funct.Mater. 19, 1399–1405. (doi:10.1002/adfm.200801824)

19. Coppola AM, Thakre PR, Sottos NR, White SR. 2014Tensile properties and damage evolution in vascular3D woven glass/epoxy composites. Compos. A Appl.Sci. Manuf. 59, 9–17. (doi:10.1016/j.compositesa.2013.12.006)

20. Gergely RCR et al. 2014 Multidimensionalvascularized polymers using degradable sacrificialtemplates. Adv. Funct. Mater. 25, 1043–1052.(doi:10.1002/adfm.201403670)

21. Dong HD, Esser-Kahn AP, Thakre PR, Patrick JF,Sottos NR, White SR, Moore JS. 2012 Chemicaltreatment of poly(lactic acid) fibers to enhancethe rate of thermal depolymerization. ACS Appl.Mater. Interface 4, 503–509. (doi:10.1021/am2010042)

22. Esser-Kahn AP, Thakre PR, Dong H, Patrick JF,Vlasko-Vlasov VK, Sottos NR, Moore JS, White SR.2011 Three-dimensional microvascularfiber-reinforced composites. Adv. Mater. 23,3654–3658. (doi:10.1002/adma.201100933)

23. Trask RS, Bond IP. 2006 Biomimetic self-healing ofadvanced composite structures using hollow glassfibres. Smart Mater. Struct. 15, 704–710.(doi:10.1088/0964-1726/15/3/005)

24. Pang JWC, Bond IP. 2005 ‘Bleeding composites’—damage detection and self-repair using abiomimetic approach. Compos. A Appl. Sci. Manuf.36, 183–188. (doi:10.1016/S1359-835X(04)00166-6)

25. Huang J-H, Kim J, Agrawal N, Sudarsan AP, MaximJE, Jayaraman A, Ugaz VM. 2009 Rapid fabricationof bio-inspired 3D microfluidic vascular networks.Adv. Mater. 21, 3567–3571. (doi:10.1002/adma.200900584)

26. Lim D, Kamotani Y, Cho B, Mazumder J, TakayamaS. 2003 Fabrication of microfluidic mixers andartificial vasculatures using a high-brightnessdiode-pumped Nd : YAG laser direct writemethod. Lab. Chip 3, 318–323. (doi:10.1039/B308452C)

27. Liu D, Yang Z, Wang P, Li F, Wang D, He D. 2013Preparation of 3D nanoporous copper-supportedcuprous oxide for high-performance lithium ionbattery anodes. Nanoscale 5, 1917–1921.(doi:10.1039/c2nr33383j)

28. Xu J, Ji X, ZhangW, Liu G. 2008 Pool boiling heattransfer of ultra-light copper foamwith open cells.

Int. J. Multiphase Flow 34, 1008–1022. (doi:10.1016/j.ijmultiphaseflow.2008.05.003)

29. Xie Z, Ikeda T, Okuda Y, Nakajima H. 2004 Soundabsorption characteristics of lotus-type porouscopper fabricated by unidirectional solidification.Mater. Sci. Eng. A 386, 390–395. (doi:10.1016/S0921-5093(04)00966-9)

30. Ding W, Pengcheng S, Changhong L, Wei W,Shoushan F. 2008 Highly oriented carbon nanotubepapers made of aligned carbon nanotubes.Nanotechnology 19, 075609. (doi:10.1088/0957-4484/19/35/355708)

31. Chu HT, Zhang ZC, Liu YJ, Leng JS. 2014 Self-heatingfiber reinforced polymer composite usingmeso/macropore carbon nanotube paper and itsapplication in deicing. Carbon 66, 154–163.(doi:10.1016/j.carbon.2013.08.053)

32. Hashemi S, Kinloch A, Williams J. 1989 Correctionsneeded in double-cantilever beam tests forassessing the interlaminar failure offibre-composites. J. Mater. Sci. Lett. 8, 125–129.(doi:10.1007/BF00730701)

33. Kousourakis A, Mouritz AP. 2010 The effect ofself-healing hollow fibres on the mechanicalproperties of polymer composites. Smart Mater.Struct. 19, 085021. (doi:10.1088/0964-1726/19/8/085021)

34. Trask RS, Williams GJ, Bond IP. 2007 Bioinspiredself-healing of advanced composite structuresusing hollow glass fibres. J. R. Soc. Interface 4,363–371. (doi:10.1098/rsif.2006.0194)

35. Wang Y, Deng S, Huang J, Pham DT, Ji C, Butt H.2015 Recovery of flexural strength in vascularself-healing composites. Internal Report, School ofMechanical Engineering, University of Birmingham,UK.

36. Aragón AM, Hansen CJ, WuW, Geubelle PH, Lewis J,White SR. 2007 Computational design andoptimization of a biomimetic self-healing/coolingcomposite material. In The 14th Int. Symp. on SmartStructures and Materials & Nondestructive Evaluationand Health Monitoring, p. 65261G. San Diego, CA:International Society for Optics and Photonics.

37. Soghrati S, Aragon AM, Geubelle PH. 2014 Design ofactively-cooled microvascular materials: a geneticalgorithm inspired network optimization. Struct.Multidiscip. Optim. 49, 643–655. (doi:10.1007/s00158-013-1000-z)

38. Cetkin E, Lorente S, Bejan A. 2011 Vascularization forcooling and mechanical strength. Int. J. Heat MassTransf. 54, 2774–2781. (doi:10.1016/j.ijheatmasstransfer.2011.02.061)

on May 22, 2018http://rsos.royalsocietypublishing.org/Downloaded from